Skip to content

[OpenMP] Fix distributed barrier hang for OMP_WAIT_POLICY=passive #83058

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 2 commits into from
Feb 27, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
5 changes: 2 additions & 3 deletions openmp/runtime/src/kmp_runtime.cpp
Original file line number Diff line number Diff line change
Expand Up @@ -5708,9 +5708,8 @@ void __kmp_free_team(kmp_root_t *root,
}
#endif
// first check if thread is sleeping
kmp_flag_64<> fl(&th->th.th_bar[bs_forkjoin_barrier].bb.b_go, th);
if (fl.is_sleeping())
fl.resume(__kmp_gtid_from_thread(th));
if (th->th.th_sleep_loc)
__kmp_null_resume_wrapper(th);
KMP_CPU_PAUSE();
}
}
Expand Down
37 changes: 37 additions & 0 deletions openmp/runtime/test/barrier/llvm-issue-80664.c
Original file line number Diff line number Diff line change
@@ -0,0 +1,37 @@
// RUN: %libomp-compile
// RUN: env OMP_WAIT_POLICY=passive \
// RUN: KMP_FORKJOIN_BARRIER_PATTERN='linear,linear' %libomp-run
// RUN: env OMP_WAIT_POLICY=passive \
// RUN: KMP_FORKJOIN_BARRIER_PATTERN='tree,tree' %libomp-run
// RUN: env OMP_WAIT_POLICY=passive \
// RUN: KMP_FORKJOIN_BARRIER_PATTERN='hyper,hyper' %libomp-run
// RUN: env OMP_WAIT_POLICY=passive \
// RUN: KMP_FORKJOIN_BARRIER_PATTERN='dist,dist' %libomp-run
//
// LLVM ISSUE 80664: https://github.com/llvm/llvm-project/issues/80664
//
// Distributed barrier + OMP_WAIT_POLICY=passive hangs in library termination
// Reason: the resume logic in __kmp_free_team() was faulty and, when checking
// for sleep status, didn't look at correct location for distributed barrier.

#include <stdio.h>
#include <stdlib.h>

int a = 0;

void test_omp_barrier() {
#pragma omp parallel
{
#pragma omp task
{
#pragma omp atomic
a++;
}
}
}

int main() {
test_omp_barrier();
printf("a = %d\n", a);
return EXIT_SUCCESS;
}